With the ever-increasing need for computer networks and the rapid development of computer-based technologies, market demand for server systems is rising. Some large companies use hundreds or even thousands of server units (also known as “blade servers”) to accommodate a significant amount of network activity. Server units may include CPUs, RAM, and/or hard drives, as well as entire servers designed to fit on small plug-and-play cards or boards. Modern data centers oftentimes house large numbers of these server units, typically in server racks.
Further, the server racks containing the server units are typically stored in a closely spaced relation in the data centers so that server units can share power supplies, monitors, keyboards, and other electronics. Storing server units close to one another also allows technicians to service numerous server units at once. In operation, however, the densely arranged server units generate a substantial amount of heat, which must be removed to prevent overheating and malfunction of the server units and other equipment.
In one configuration known in the art, the server racks are placed in a server room of a data center with a central ventilation system that provides cooling air at least between separate rows of server racks. The drawback with this type of configuration is that the cooling air is not directed specifically at the server rack, which in turn decreases the cooling effect of the ventilation system. Thus, to maintain the server units at a desirable temperature, the system typically consumes an inordinate amount of energy.
Another cooling configuration known in the art involves fans and/or cooling systems arranged to blow air through ventilation holes in the front door of the server rack. The problem with this configuration, though, is that airflow through the server racks is usually poor. Components within the server racks, such as a multitude of cables, power strips extending orthogonally from the sidewalls of the server racks, and the server units themselves, for example, can impede the flow of air through the server rack.
Still another problem with existing server racks concerns serviceability. Technicians are required to service the contents of a server rack on occasion, such as by replacing server units, rewiring or rerouting electrical cables, or troubleshooting, for instance. However, because server racks most often contain a multitude of cables, access to any given server unit may be partially or entirely obstructed by the cables. As a result, servers that do not require servicing must be temporarily taken offline while other server units are serviced.
Another feature of existing server racks that contributes to poor serviceability and poor airflow is the structure of the power strips. Many server racks include a three-phase power strip (e.g., a three-phase power distribution unit (PDU)), as many server units nowadays require power from combinations of power phases. Current three-phase power strips typically have an A-phase at the bottom, a B-phase above the A-phase, and a C-phase at the top. Thus some server units towards the top of the server rack may require an electrical connection to the A-phase of the power strip amongst other connections, and some server units towards the top of the server rack may require at least an electrical connection to the C-phase of the power strip amongst other connections. Understandably, cables quickly become crisscrossed, further limiting access to server units within the server rack.
For at least the forgoing reasons, there is a demonstrated need to improve the cooling efficiency and serviceability of server racks.